.D. Thesis Defense Announcement

Methods for Enriching Transportation Survey Datasets: With Sample Applications Using Psychometric Variables

by

Faaiqa Atiyya Shaw

Advisor(s):

Dr. Patricia L. Mokhtarian (CEE)

Committee Members:

Dr. Michael P. Hunter (CEE), Dr. James S. Roberts (PSYC),

Dr. Giovanni Circella (CEE), Dr. Ram Pendyala (SSEBE, ASU), Dr. Kari E. Watkins (CEE)

Date & Time: January 27, 2021, 11 a.m. EST

Location: https://bit.ly/3siS3Z8

 Complete announcement, with abstract, is attached

Technological disruptions, environmental and health upheavals, and societal shifts are just a few of the major forces interacting in rapid and unprecedented ways to influence how we live, work, and navigate within our built environments. Transportation engineers and urban planners must grapple with how such widespread, and in many cases, yet unknown, changes will alter the urban landscape, shifting travel patterns and requiring a fresh look at infrastructure forecasting, planning, and development into the future. In a time of such uncertainty, it is increasingly important for national, state, and regional planning organizations to be able to understand and forecast behavioral and attitudinal changes. However, modeling such shifts depends on actively collected survey data, which are infrequently gathered, time and cost-intensive, and suffer from continuously declining response rates (and accompanying biases). 
 
Accordingly, the work presented in this thesis aims to address some of these challenges by making use of data driven tools like machine learning within the context of the rapidly growing big data landscape to develop and present three approaches for supplementing and/or expanding transportation survey datasets using active and passive data streams. Broadly, these approaches include: (1) exploring and utilizing novel sources of data for transportation modeling and analysis; (2) integrating and expanding existing, traditional sources of transportation data with both active and passive data sources; and (3) developing marker statements for expanding the breadth of information obtained without substantially increasing survey lengths. These methods are demonstrated by applying them to enrich traditional transportation surveys with psychometric data (e.g., attitudes), which have been shown in the literature to have the ability to explain and predict behaviors, but which are often not captured on surveys. 
 
Each application is validated by integrating the enriched datasets within sample travel behavior models, and observing changes in predictive accuracy, model fit, and interpretability. Findings show that expanded variable richness for transport surveys, specifically with psychometric variables like attitudes, can improve performance and interpretability of travel behavior models. This research has societal implications that center on the potential for improved travel demand forecasting and behavioral predictions. Such improvements can facilitate more efficient expenditures, improve infrastructure planning, and ultimately increase quality of life for all. Even more broadly, the methods of this research may be applied to enrich many more large-scale behavior-based surveys with diverse variables, thereby providing richer, more robust data streams for use in an array of modeling and forecasting efforts.